论文标题
通过生成对抗网络快速生成大规模结构密度图
Fast Generation of Large-scale Structure Density Maps via Generative Adversarial Networks
论文作者
论文摘要
生成对抗网络（GAN）是无监督的机器学习的最新进步。它们是两个神经网络之间的猫鼠游戏：[1]与培训集相比，学会验证样本是真实还是假货的歧视网络，而[2]生成器网络学会生成似乎属于培训集的数据。两个网络互相学习，直到完成培训，并且发电机网络能够生成与培训集无法区分的样本。我们发现，甘恩非常适合快速生成新型的3D密度图，这些密度与从N体模拟获得的图形无法区分。在几秒钟内，训练有素的甘恩可以在宇宙历史上的不同时期生成数千个密度图。然后可以使用这些gan生成的地图来研究大规模结构的演变。
Generative Adversarial Networks (GANs) are a recent advancement in unsupervised machine learning. They are a cat-and-mouse game between two neural networks: [1] a discriminator network which learns to validate whether a sample is real or fake compared to a training set and [2] a generator network which learns to generate data that appear to belong to the training set. Both networks learn from each other until training is complete and the generator network is able to produce samples that are indistinguishable from the training set. We find that GANs are well-suited for fast generation of novel 3D density maps that are indistinguishable from those obtained from N-body simulations. In a matter of seconds, a fully trained GAN can generate thousands of density maps at different epochs in the history of the universe. These GAN-generated maps can then be used to study the evolution of large-scale structure over time.
